This project equips AI agents, particularly those using the Model Context Protocol (MCP) like Claude and Cursor, with autonomous email capabilities. It solves integration challenges by providing a free, no-signup service with 27 email tools. Agents can send, receive, manage contacts, and automate email tasks without human intervention or complex API key management, enabling seamless AI-driven communication.

How It Works

The agent-kit acts as an MCP server, exposing 27 email tools via JSON-RPC over stdio, powered by KeyID.ai's free email infrastructure. This abstracts email protocol complexities, allowing AI agents to fetch inboxes, send scheduled emails, reply, manage contacts, and configure auto-replies, facilitating autonomous AI email handling and simplifying agent development.

Quick Start & Requirements

Install via npm (npm install @keyid/agent-kit) or yarn (yarn add @keyid/agent-kit). For Claude Desktop, configure claude_desktop_config.json under mcpServers.keyid with command: "npx" and args: ["@keyid/agent-kit"]. Environment variables KEYID_PUBLIC_KEY and KEYID_PRIVATE_KEY (hex) are required for persistent identity, though auto-generation is supported. Alternatively, run directly: npx @keyid/agent-kit or KEYID_PUBLIC_KEY=abc123 KEYID_PRIVATE_KEY=def456 npx @keyid/agent-kit. Prerequisites include Node.js and an MCP client.

Highlighted Details

• Provides 27 MCP tools for comprehensive email management: inbox fetching (search/filter), sending (scheduled, HTML, display name), replying, forwarding, message updates, unread counts, threads, drafts, auto-reply, signatures, contacts, webhooks, and list management.
• Leverages KeyID.ai's free email infrastructure specifically for AI agents.
• Requires no human registration, API keys, or cost.
• Enables autonomous AI email handling, including scheduled delivery and vacation responders.
• Uses MCP protocol version 2024-11-05.

Maintenance & Community

The README provides no specific details on maintainers, community channels (Discord, Slack), roadmaps, or sponsorships.

Licensing & Compatibility

Licensed under MIT, permitting commercial use, modification, distribution, and closed-source linking. This offers broad compatibility for integration.

Limitations & Caveats

Designed exclusively for AI agents and requires an MCP client. Persistent agent identity relies on managing Ed25519 keypairs, though auto-generation is available. The README does not detail performance benchmarks or known bugs.